Gradient Descent (10)
Michael L. Littman
February 16th, 1999
NEURAL NETWORKS
What's a Neural Network
Supervised Learning
Nonincremental vs. Incremental
Neural Networks
Network Function Approximators
Basic ``Neuron''
Linear Activation Function
Activation Functions
Feed-Forward Networks
Recurrent Networks
Computational Power
Perceptron Behavior
Gradient Descent Idea
Perceptron Training Rule
Gradient Descent on Squared Error
Perceptron Learning: Good
Perceptron Learning: Bad
Linear Separability
Example
MULTILAYER PERCEPTRONS
What Does This Buy Us?
Backpropagation
Backpropagation's Claim to Fame
Problems
Overfitting
Cross Validation Idea
Potential Methods for Improving Performance
Next:
NEURAL NETWORKS
